Referral outcomes of attendances at general practitioner-led urgent care centres in London, England: retrospective analysis ofhospital administrative data
Thomas E Cowling,1 Farzan Ramzan,1 Tim Ladbrooke,2 Hugh Millington,3 Azeem Majeed,1 Shamini Gnani1
1Department of Primary Care and Public Health, Imperial College London, London, UK
2London Central and West Unscheduled Care Collaborative, London, UK
3Emergency Department, Charing Cross Hospital, Imperial College Healthcare NHS Trust, London, UK
Correspondence to:
Thomas E Cowling
Imperial College London
Department of Primary Care and Public Health
Reynolds Building, St Dunstan’s Road
London W6 8RP
United Kingdom
Tel: +44 (0)20 7594 0779
Keywords: Emergency Medical Services; Hospital Emergency Services; Urgent Care Centers; General Practitioner; Primary Care Physician
Word count: 3076
Abstract
ObjectiveTo identify patient and attendance characteristics that are associated with onwards referral to co-located emergency departments (EDs) or other hospital specialty departments from general practitioner-led urgent care centres (UCCs) in northwest London, England.
MethodsWe conducted a retrospective analysis of administrative data recorded in the UCCs at Charing Cross and Hammersmith Hospitals, in northwest London, fromOctober 2009 toDecember 2012. Attendances made by adultsresident in England were included. Logistic regression was used to model the associations between the explanatory variables – age; sex; ethnicity; socioeconomic status; area of residence; distance to UCC; general practitioner registration; time, day, quarter, year, and UCC of attendance – and the outcome of onwards referral to the co-located EDs or other hospital specialty departments.
ResultsOf 243 042 included attendances,74.1% were managed solely within the UCCs; without same-day referral to the EDs (16.8%) or other hospital specialty departments (5.7%), or deferred referral to a fracture, hand management, or soft tissue injury management clinic (3.3%). The adjusted odds of onwards referral was estimated to increase by 19% (OR 1.19, 95% CI 1.18 to 1.19) for a ten year increase in a patient’s age. Males, patients registered with a GP, and residents of less socioeconomically deprived areas were also more likely to be referred onwards from the UCCs.
ConclusionsThe majority of patients, across each category of all explanatory variables, were managed solely within the UCCs, though a large absolute number of patients were referred onwards each year.Several characteristics of patients and their attendances were associated with the outcome variable.
Introduction
Urgent care centres (UCCs) vary widely in form, both between and within countries. In the US, a service is consideredan urgent care centre if it: primarily provides walk-in care; is open every weekday evening and at least one weekend day; and provides onsite x-rays and suturing for minor lacerations.1 The majority of these centres are staffed by physicians, commonly specialising in family practice, and nurses and physician assistants; most centres can provide fracture care and intravenous fluids.1 In England, the health service regulator considers urgent care centres to: be led by family practice physicians (general practitioners, GPs); provide extended opening hours or be open 24 hours a day, seven days a week; and provide urgent care on a walk-in basis, but not necessarily to patients with ‘non-urgent’ needs.2 Not all English UCCs have access to x-ray facilities or provide fracture care,3whilst only some are located alongside hospital emergency departments (EDs).
The UK Royal College of Emergency Medicine acknowledges both the significant variation in UCC design and system integration, and the lack of research on the role of UCCs.4 Despite this, it proposes that UCCsshould be co-located with EDs to prevent patients without a clinical need for ED care from contributing to ED workload,4particularly since a considerable number of visits to EDs each year could likely be managed by a general practitioner (GP).5-7In other European countries, such as the Netherlands8 and Belgium,9 primary care cooperatives have been integrated with EDs, often for the same reasons as those given in England. In the US, whilst UCCs are typically independent physician offices, a considerable percentage of ED visits have been estimated to be treatable in UCCs.10
In northwest London, England, UCCsstaffed by GPs and emergency nurse practitioners were first co-located with the EDs at Charing Cross and Hammersmith Hospitals in 2009.11 The centres were intended to reduce rates of ED visits and short-stay emergency admissions to hospital in particular. National policy also seeks to achieve this aim;12the annual number of visits to accident and emergency (A&E) departments and emergency hospital admissionsincreased from 19.6 million to 21.7 million and 5.0 million to 5.3 million from 2008-09 to 2012-13 respectively.13 14The ED at Hammersmith Hospital closed on 10th September 2014, as part of plans to centralise ED services in northwest London.15 16The future of the ED at Charing Cross Hospital is uncertain,15 17but the UCCs are likely to remain at both hospitals.
The objective of this analysis was to identify patient and attendance characteristics that are associated with onwards referral from the UCCs to the co-located EDs or to other hospital specialty departments (versus management solely within the UCCs) at Charing Cross and Hammersmith Hospitals.
Methods
Study design, setting, and population
We conducted a retrospective, cross-sectional analysis of attendance data recorded in two GP-led UCCs in northwest London, England; Fulham and Hammersmith Centres for Health are located at Charing Cross Hospital and Hammersmith Hospital respectively. The data period was 1 October 2009 to 31 December 2012.
During this period, both UCCswereon the same site as consultant-led EDsthat provided 24-hour services. The UCCs provided ‘open access’ urgent care services, for which an appointment was not necessary, seven days a week. Fulham UCC provided thisservice 24 hours a day, whilst Hammersmith UCC provided it from 8am to 10.30pm.
Patients visiting the UCCs first registered with reception and were then allocated to a specific treatment stream by a GP (see Appendix 1). The streaming decision, taking approximately three minutes, determined whether a patient was immediately referred to the emergency department or initially managed in the UCC. This decision was underpinned by evidence based guidelines that were developed by a multi-disciplinary team including: GPs; emergency medicine consultants; gynaecologists; psychiatrists; and emergency nurse practitioners. The principles adopted in the development of the guidelines were: the UCCs would see any patient that a GP would see in their ‘normal surgery’ and would only refer to the EDs those patients that required an urgent opinion, as in routine general practice. The streaming guidelines evolved over time, through discussion of clinical cases and audits of ED referrals, and included updates of other guidelines such as those of the National Institute for Health and Care Excellence (NICE).18
The six streams were:
1. Emergency Department – a patient requires facilities present in the emergency department;
2. GP Priority – a patient is seen by a GP;
3. Minor Illness – a patient is seen by a GP or emergency nurse practitioner;
4. Minor Injuries – a patient is seen by an emergency nurse practitioner;
5. See and Treat – a patient is seen and treated by the GP Streamer; and
6. Reception Navigation – a patient is referred to another service by reception staff.
Patients could not access the emergency departments without first seeing the GP Streamer in the UCCs unless they arrived by ambulance or had already been seen by their own GP who referred them to hospital. Further details on the service model have been described elsewhere.11 19 20
Our dataset included all walk-in patients atthe UCCs over the study period, in addition to patients who had arrived by ambulance to the EDs and were streamed to the UCCs by an ED nurse. Patients who had arrived by ambulance and were deemed to require ED servicesimmediately were not included in the study, as they were not seen in the UCCs. In addition, we excluded patients who had been referred to the UCCs by a GP (‘Expected Specialty Patients’; 1.8% of attendances) and patients aged <18 years old (7.4%), as young children are generally referred to paediatric services and attendances for adolescent patientshave been analysed previously.20 Patients who lived outside of England were also excluded (0.3%), due to missing data.
Variables
The outcome variable was onwards referral from the UCCs to the EDs or other hospital specialty departments, as opposed to management of an attendance solely within the UCCs.
The explanatory variables included patientcharacteristics – age, sex, ethnicity, socioeconomic status, area of residence, distance from residence to UCC, and GP registration status – theUCC attended, and the time, day, quarter, and year of attendance. Age, sex, ethnicity, GP registration status, and the time/date of attendance are routinely recorded in the administrative database used in the UCCs. Socioeconomic status was measured using the Index of Multiple Deprivation (IMD) rank assigned to the Lower Layer Super Output Area in which patients resided.21Area of residence was derived from patients’ postcodes, as were the straight-line distances between patients’ residences and the UCCs they attended. Records with missing data for one or more variables (5.2%) were excluded.
Patient diagnoses were recorded in the UCC administrative datasets as Read codes—the classification system widely used in the computer systems of UK general practices. We described the most common primary diagnoses (recorded in the first of six fields) but did not include this data in further analysis due to missing data (19% of attendances had no primary diagnosis recorded).
Statistical methods
We estimated the association between the log odds of a patient being referred to the EDs or other hospital specialty departmentsand the explanatory variables using logistic regression. All explanatory variables were entered into the multivariable regression model, asalmost all variables were statistically significant in the univariable analysis. We recoded continuous variables into categorical variables for the regression analyses to facilitate interpretation and, for the variable related to distance to the UCC attended, to account for non-linear associations. We did not test for interactions due to the large number of explanatory variables in the model and the absence of strong hypotheses specified a priori concerning expected interactions. The link test indicated that the model was specified correctly, and the percentage of standardised Pearson residuals within ±2 was 96.3%, suggesting good model fit.Data processing and analysis were conducted using Stata MP Version 13.1 (StataCorp, College Station, TX, USA).
Results
Description of attendances
The analysis included 243042 attendances to Fulham and Hammersmith Centres for Health between 1 October 2009 and 31 December 2012.Most patients were managed in the UCCs (74.1% of attendances) without requiring same day referralto the EDs (16.8%). A further 5.7% of attendances were referred to be seen by a hospital specialist on the same day,and 3.3% required a deferred referral to a fracture, hand management, or soft tissue injury management clinic (table 1). The flow of attendances through the UCCs is shown in figure 1.
The majority of attendances (64.2%) were at Fulham Centre for Health. The annual number of attendances increased by 8.9% from 73 083 in 2010 to 79 590 in 2012. The percentage of attendances managed within the UCCs was relatively constant over this period (figure 2). Most attendees(86.0%) were registered with a general practice, and half (48.7%) of attendees were residents of the London Borough of Hammersmith and Fulham.
Fracture of upper limb was the commonest primary diagnosis recorded for patients referred on from the UCCs (11.8% of attendances referred on). This was followed by sprains and strains not otherwise specified (11.1%), abdominal pain (6.2%), and fracture of lower limb (5.7%)(table 2).
Associations with patient characteristics
Older patients were more likely to be referred to the EDs or other hospital specialty departments; the percentage of patients ≥80 years of age experiencing this outcome was approximately double that for patients aged 18-29 years old (41.3% versus 20.4%,OR2.74, 95% CI 2.59 to 2.91) (table 3). Younger patients were less likely to be registered with a GP (≥80 years versus 18-29 years, 95.4% versus 79.5%,OR 1.67, 95% CI1.54 to 1.80) and patients who were not registered with a GP were less likely to be referred onwards (20.3% versus 26.8%, OR0.70, 95% CI0.68 to 0.72).
In the multivariable analysis, which controlled for the associations between explanatory variables, similar results to those given above were obtained (table 3).A ten year increase in patient age was associated with a 19% greater odds of onwards referral (OR 1.19, 95% CI 1.18 to 1.19).Onwards referral was also estimated to be more likely for malesthan females in the multivariable analysis (OR: 1.11, 95% CI1.09 to 1.13), as it was for patients with higher socioeconomic status (least deprived fifth relative to most deprived fifth, OR 1.15, 95% CI1.09 to 1.22).
Patients resident in London boroughs other than Hammersmith and Fulham were more likely to be referred onwards (relative to Hammersmith and Fulham residents, 27.8% versus 24.4%,OR 1.19, 95% CI 1.17 to 1.22). However, areaof residence was not associated with the outcome variable after controlling for the distance between patients’ residences and the UCCs they attended. Increased distance from the UCC was generally associated with greater adjusted odds of referral to the EDs or other hospital specialty departments (5.0-9.9km relative to <1km, OR 1.40, 95% CI 1.34 to 1.47), but patients residing ≥30km from the UCC attended were less likely to be referred onwards than those living <1km from the UCC (OR 0.85, 95% CI 0.76 to 0.99).
Associations with attendance characteristics
The percentage of patients referred to the EDs or other hospital specialty departments was fairly consistent between the two UCCs and across time periods, days of the week, quarters of the year, and years (table 4). This percentage was greatest between the times of 01:00 and 04:59 (32.4%), on Fridays (27.5%), and from April to June (26.8%), although most patients could still be managed within the UCCs during these periods.Patients attending the UCCs in 2010 (27.1%, OR 1.16, 95% CI 1.11 to 1.22) and 2011 (26.0%, OR 1.15, 95% CI 1.11 to 1.20) were more likely to be referred onwards than patients in 2009 (23.4%),adjusting for all other explanatory variables.
Discussion
Summary of findings
The majority of included visits to the UCCs at Charing Cross and Hammersmith Hospitals were managed solely within the UCCs. This finding was consistent across each category of all variables examined in the analysis.The age of patients had the largest association with the odds of onwards referral to the co-located EDs or other hospital specialty departments. Yet the majority (59.7%) of patients ≥80 years of age were still managed solely within the UCCs. Even amongst patients visiting the UCCs in the early hours of the morning, most visits were managed within the UCCs (67.6% of visits between 01:00 and 04:59).
The main diagnoses referred from the UCCs to the EDs or other hospital specialty departments were musculoskeletal; fractures of the upper or lower limb, sprains, and strains accounted for 28.6% of referrals. Most referrals for sprains and strains were made to the ED soft tissue and minor fracture clinics, and a small number were made to the orthopaedic-run fracture clinic. While there may be potential to improve the percentage of patients referred on by the UCCs, we are limited in this study in using the main Read code diagnosis to determine appropriateness, without reference to clinical records and management by the fracture clinics.
Abdominal pain was also frequently referred on (6.2% of onwards referrals), to the duty general surgeons. We are uncertain whether access to abdominal ultrasound scans would improve the UCCs’ capacity to manage more patients without onwards referral. These scans could be ordered through the hospital imaging department (as for X-rays) or GPs could be trained to use ultrasound scanners in the UCCs. The relatively large percentage of elderly patients referred onwards suggests that it might be helpful for doctors specialising in care of the elderly to work alongside GPs and emergency nurse practitioners in the UCCs. The cost-effectiveness of these changes would need to be evaluated.
Our findings are likely to be particularly affected bythe streaming guidelines used by the GPs and may evolve over time with changes in referral behaviour. However,the percentage of attendances managed solely within the UCCs has remained relatively constant since their introduction thus far.
Relation to existing literature
We have previously reported why patients with minor illness choose to attend the UCCs and patterns of use among adolescent attendees.19 20 To our knowledge, no other studies have examined GP-led UCCs co-located with hospital EDs in England.
One study has evaluated two independent GP-led walk-in centres in England, using a questionnaire survey and analysis of routine data.22 23 Respondents commonly stated that the superior access provided by the centres relative to their general practices was a reason for their visit,22 and some would have visited an ED if the walk-in centres did not exist.23A controlled before-and-after study of eight hospital EDs with co-located nurse-led walk-in centres concluded that there was no evidence of an effect on attendance rates, costs, or outcomes of care;24 yet the centres were only implemented to a limited extent24and patients were frequently unable to distinguish between the ED and the co-located walk-in centres.25Other studies have examined independent nurse-led walk-in centres in England with no evidence of an effect on neighbouring services.26 27
EDs in Switzerland have also implemented co-located GP-led services to manage patients with less urgent problems, and achieve cost savings.28 Several studies in the Netherlands29-31and one in Ireland32, mainly using before-and-after designs, report a reduction in ED utilisation with the introduction of out-of-hours GP cooperatives within or near to EDs. Similar studies in England33and Belgium34did not observe this effect.A Cochrane review of evidence on the effect of GPs providing care for non-urgent patients within EDs concluded that there was insufficient evidence with which to draw conclusions.35
Strengths and limitations
The data period started with the first month that both UCCs were operating, October 2009, and extended over a 39 month period, to December 2012. Hence, a comprehensive set of attendances were analysed and seasonal and year trends could be accounted for. The dataset would also have captured all hospital-based urgent care in the London Borough of Hammersmith and Fulham over this period, as Charing Cross and Hammersmith Hospitals are the only hospitals in this area.